Cross validation and regularization


I just finished the “Regularization and bias/variance” lecture video. So far, I have an impression that cross-validation helps avoid both high bias and high variance, while regularization helps avoid only high variance. Is my understanding correct?

Does using cross-validation with regularized cost function focus more on avoiding high variance than on avoiding high bias?

Thank you!

Hello @chaochun,

Thank you for the question! Let me jump directly to a slight difference between cross-validation and regularization. The former helps you discover a high bias / high variance problem, and given the discovery is a high variance problem, you may apply the later to literally helps you reduce variance.


Have a great day.

1 Like